Scoring system for mortality prediction of in-hospital COVID-19 patients in resource-limited settings: a single center cohort study during Delta and Omicron waves

Phey Liana; Krisna Murti; Iche Andriyani Liberty; Zen Hafy; Tungki Pratama Umar

doi:10.18502/fem.v7i4.14475

Phey Liana Clinical Pathology Department, Faculty of Medicine, Universitas Sriwijaya-Dr. Mohammad Hoesin Hospital, Palembang, Indonesia.
Krisna Murti Anatomic Pathology Department, Faculty of Medicine, Universitas Sriwijaya-Dr. Mohammad Hoesin Hospital, Palembang, Indonesia.
Iche Andriyani Liberty Public Health Department, Faculty of Medicine, Universitas Sriwijaya, Palembang, Indonesia.
Zen Hafy Biomedical Department, Faculty of Medicine, Universitas Sriwijaya, Palembang, Indonesia.
Tungki Pratama Umar UCL Centre for Nanotechnology and Regenerative Medicine, Division of Surgery and Interventional Science, University College London, London, UK.

DOI: https://doi.org/10.18502/fem.v7i4.14475

Keywords: COVID-19; Prognostic; Scoring System; Survival

Abstract

Objective: Coronavirus Disease 2019 (COVID-19)-related mortality includes several risk variables that are country-specific in nature. The development of a scoring system is necessary regarding the appearance of novel virus variants. The objective of this research is to develop a prognostic score for COVID-19 patients in resource-constrained settings.

Methods: This study used a retrospective and prospective cohort design to identify variables that influence COVID-19 patients' in-hospital mortality. The receiver operating characteristic (ROC) curve analysis was utilized to determine the laboratory variables cut-off. Cox regression analysis was undertaken to determine the exact variables influencing the survival of COVID-19 patients. A scoring system was created using the best model based on the Hosmer-Lemeshow test (calibration) and the area under the curve (AUC) (discrimination ability).

Results: Based on calibration and discrimination testing, model 2 (immune disorders, unconsciousness, cerebrovascular disease, onset, and oxygen saturation) was rated as the most advantageous model. Model 2 (without age adjustment) had a superior AUC than model 2A (with age). Cut-off was determined at 2, and calculated for onset ≥7 days (AUC=0.816, 95% CI: 0.742,0.890) and <7 days (AUC=0.850, 95% CI: 0.784,0.916). There was no difference in scoring system utilization for subjects recruited during Delta or Omicron waves (P=0.527).

Conclusion: The model (cut-off value ≥2) which incorporated age ≥65 years, immune disorders, decreased consciousness, increased respiratory rate, and oxygen saturation <95% is the best model in our study to predict COVID-19 patient mortality.